"""Configuration loading for DTD Pipeline."""

import json

import os

from pathlib import Path

from typing import Optional

from pydantic import BaseModel, Field

from models import PodcastConfig

class Config(BaseModel):

"""Pipeline configuration loaded from environment."""

# Taddy API

taddy_api_key: str = Field(default="")

taddy_user_id: str = Field(default="")

lookback_days: int = Field(default=7)

# Whisper

whisper_model: str = Field(default="large-v3")

whisper_device: str = Field(default="cuda")

whisper_compute_type: str = Field(default="float16")

# Pyannote

hf_token: str = Field(default="")

pyannote_model: str = Field(default="pyannote/speaker-diarization-3.1")

# Speaker ID

speaker_match_threshold: float = Field(default=0.70)

# Supabase

supabase_url: str = Field(default="")

supabase_key: str = Field(default="")

supabase_bucket: str = Field(default="podcast-transcripts")

# Paths

data_dir: Path = Field(default=Path("data"))

audio_dir: Path = Field(default=Path("data/audio"))

speakers_dir: Path = Field(default=Path("data/speakers"))

logs_dir: Path = Field(default=Path("data/logs"))

state_db_path: Path = Field(default=Path("data/state.db"))

# Podcasts config

podcasts_config_path: Path = Field(default=Path("podcasts.json"))

def load_config() -> Config:

"""Load configuration from environment variables."""

return Config(

# Taddy

taddy_api_key=os.getenv("TADDY_API_KEY", ""),

taddy_user_id=os.getenv("TADDY_USER_ID", ""),

lookback_days=int(os.getenv("LOOKBACK_DAYS", "7")),

# Whisper

whisper_model=os.getenv("WHISPER_MODEL", "large-v3"),

whisper_device=os.getenv("WHISPER_DEVICE", "cuda"),

whisper_compute_type=os.getenv("WHISPER_COMPUTE_TYPE", "float16"),

# Pyannote

hf_token=os.getenv("HF_TOKEN", ""),

pyannote_model=os.getenv("PYANNOTE_MODEL", "pyannote/speaker-diarization-3.1"),

# Speaker ID

speaker_match_threshold=float(os.getenv("SPEAKER_MATCH_THRESHOLD", "0.70")),

# Supabase

supabase_url=os.getenv("SUPABASE_URL", ""),

supabase_key=os.getenv("SUPABASE_KEY", ""),

supabase_bucket=os.getenv("SUPABASE_BUCKET", "podcast-transcripts"),

# Paths

data_dir=Path(os.getenv("DATA_DIR", "data")),

audio_dir=Path(os.getenv("AUDIO_DIR", "data/audio")),

speakers_dir=Path(os.getenv("SPEAKERS_DIR", "data/speakers")),

logs_dir=Path(os.getenv("LOGS_DIR", "data/logs")),

state_db_path=Path(os.getenv("STATE_DB_PATH", "data/state.db")),

podcasts_config_path=Path(os.getenv("PODCASTS_CONFIG_PATH", "podcasts.json")),

)

def load_podcasts(config: Config) -> list[PodcastConfig]:

"""Load podcast configurations from JSON file."""

config_path = config.podcasts_config_path

if not config_path.exists():

return []

with open(config_path) as f:

data = json.load(f)

return [PodcastConfig(**p) for p in data.get("podcasts", [])]

# Global config instance

_config: Optional[Config] = None

def get_config() -> Config:

"""Get or create global config instance."""

global _config

if _config is None:

_config = load_config()

return _config